Most of the political potentialities of automated surveillance depend on two elements. The debate has generally concentrated on the first: data acquisition. But “digital wiretapping”, in whatever technical shape it may come, is only one part of the issue. Making sense of the data collected is the more complicated, albeit often neglected, half of the equation. Data mining technologies have certainly much advanced over the last couple of years but the commercial applications are not necessarily adapted to the demands that government agencies might have. This article makes the claim that scientist have developed a software that can “help detect terrorists before they strike”. It reads:

Computer and behavioral scientists at the University at Buffalo are developing automated systems that track faces, voices, bodies and other biometrics against scientifically tested behavioral indicators to provide a numerical score of the likelihood that an individual may be about to commit a terrorist act.

Although I’m quite skeptical about the actual performance of such software in the field (the real world is pretty messy after all) it shows the direction things are heading. The actual piece of software comes pretty close to what Virilio has described as “machine de vision” (vision machine) – a device that not only records reality (camera) but also interprets it (the man on the subway platform walking around nervously is not just a heap of pixels but a potential terrorist). Virilio talks about the “delegation of perception” and this is perhaps the most interesting aspect of the increasing technologizing of control: part of the process of decision-making (and therefore part of the responsibility) is transferred to algorithms and questions of professional ethics become matters of optimizing parameters and debugging.

Post filed under metatechnologies, surveillance.

One Comment

  1. I made this distinction in a paper on visual privacy I wrote last year, arguing that image recognition, not image capture, was what privacy advocates should be focusing on. I also pointed out that the algorithms to which recognition is delegated could themselves be designed to to delegate the work of analysis to human laborers, and in doing so achieve much better performance, but without really empowering the “humans in the loop” to make ethical judgments.

Leave a Reply

Your email address will not be published. Required fields are marked *

Tech support questions will not be answered. Please refer to the FAQ of the tool.